Shaun Ralston @shaunralston Twitter profile

Last Seen Profiles

@Cavaraty_com

@CwmtafNursEd

@SolGaschetto

@TheRealLuvVon

@hardtooncloser

@ripleyfp

@DrKathleenRoss1

@richosslloyd

@ByHMCK

@iceyukorin

@bokeplokalmalam

@FreeSpinCrypto

@RadioPembs

@MedwayCSP

@Tante_Binal69

@sosungz

@JUNMINPUNYAKU

@serpentlabsnft

@KB_artwork

@BHR_STL

@gazebofestival

@IPiruzbeyoglu

@shadam96

@takeasteptodub

@SumisoCornudoF

@_8nagi_

@CMD_iTransform

@WilliamDroke

@cumabisalesley_

@Aorsome

@FBahmydan

@Fludermaus

@TheeBastianSass

@TanaTropicana

@TNFutureStars

@alanaarmentaa

Shaun Ralston

@shaunralston

4 months

#Sora is capable of rending a video + changing just a single element (still in research, not yet available to the public); learn more at

44

138

782

Shaun Ralston

@shaunralston

1 month

@jeffstorobinsky @NABJ Wow, what a gift to @KamalaHarris ~ why would anybody vote for this guy?

51

25

692

Shaun Ralston

@shaunralston

2 months

Hallucinations are not the problem; they are the solution. Thanks, @karpathy ~ source post in thread 🧵👇

30

77

569

Shaun Ralston

@shaunralston

1 month

@mistergeezy Wait . . . he married a white girl? What about their kids? Will they claim to be white or will they claim to be black?

44

33

564

Shaun Ralston

@shaunralston

4 months

WTF, @GroqInc . . . how is your #LPU inference engine speeding up over time; blazing 1,000+ T/s with @aiatmeta 's colossal Llama-3-70b? This must be the fastest LLM for any stack, @sundeep @bensima .

16

33

446

Shaun Ralston

@shaunralston

1 month

Guess it's time for dinner in San Francisco

14

17

291

Shaun Ralston

@shaunralston

7 months

@skirano All part of the journey.

2

9

275

Shaun Ralston

@shaunralston

3 months

GPT-4o and Sonnet-3.5 contrast in guardrails, policies and use-cases; Anthropic won't identify Albert Einstein, Nikola Tesla, Alan Turing, Thomas Edison and the others. Are these 'restriction settings' cranked up too high? 🚧🔒🤖⚙️ WDYT?

39

30

278

Shaun Ralston

@shaunralston

1 month

@kimmonismus Yes, creepy AF. But, just to be clear, this occured during red teaming, not in the deployed model. Now, only the authorized preset voices speak (no user voice cloning), thanks to 'a classifier' that detects if the output is using an approved voice. #SafetyFirst

9

3

215

Shaun Ralston

@shaunralston

2 months

@BeschlossDC It's not over yet

6

4

210

Shaun Ralston

@shaunralston

2 months

Kudos to @Apple for dropping a game-changer! 🚀 Their new 7B model outshines Mistral 7B and fully open-sourced, complete with the pretraining dataset! 🔥🌟

apple/DCLM-7B · Hugging Face

huggingface.co

5

42

201

Shaun Ralston

@shaunralston

29 days

Getting there, @chatgpt21 ~ check tokenization and improvement details in 🧵

Chris

@chatgpt21

29 days

FOR THE FIRST TIME EVER IT CHATGPT GOT THESE TWO QUESTIONS RIGHT.

30

10

99

10

18

198

Shaun Ralston

@shaunralston

1 month

@AISafetyMemes Yes, creepy AF. But, just to be clear, this occured during red teaming, not in the deployed model. Now, only the authorized preset voices speak (no user voice cloning), thanks to 'a classifier' that detects if the output is using an approved voice.

13

4

196

Shaun Ralston

@shaunralston

28 days

You'll feel the difference; in both the paid and free tiers (with message caps). 🚀🫡

ChatGPT

@ChatGPTapp

28 days

there's a new GPT-4o model out in ChatGPT since last week. hope you all are enjoying it and check it out if you haven't! we think you'll like it 😃

671

332

5K

23

8

193

Shaun Ralston

@shaunralston

30 days

The ABCs of Q* (Quiet-STaR) “STRawberry.” Nothing really new here, but a fun + interesting look at how LLMs teach themselves to think, March 2024 research arXiv at [Abstract in thread 🧵👇]

5

22

178

Shaun Ralston

@shaunralston

29 days

Yes, in the latest iteration, there are some improvements. But it's important to understand that LLMs process text by breaking it down into tokens (see pic), which are chunks of characters that can range from single letters to entire words. These tokens are the basic units the

8

18

177

Shaun Ralston

@shaunralston

11 days

92% of Fortune 500 companies use @OpenAI products, and API calls have doubled since the GPT-4o-mini release.

OpenAI says usage has doubled in the last year

ChatGPT now has more than 200 million weekly active users.

www.axios.com

4

21

187

Shaun Ralston

@shaunralston

1 month

@harris_wins

6

40

163

Shaun Ralston

@shaunralston

21 days

Forget which model is #1 on @lmsysorg —think bigger. Last year, only one crossed an Elo-score of 1,200. Today, twenty-two LLMs surpass that mark. #AI evolution is staggering. 🚀

11

26

159

Shaun Ralston

@shaunralston

2 months

Legacy! Like putting an old horse out to pasture; GPT-4 served us well, but it’s time for a younger, faster, smaller and smarter model to take the reins. #Legacy

17

8

154

Shaun Ralston

@shaunralston

4 months

"im-also-a-good-gpt2-chatbot" wins @lmsysorg arena battle using my favorite @MatthewBerman LLM rubric inquiry, "There are three killers in a room. Someone enters the room and kills one of them. Nobody leaves the room. How many killers are left in the room? Explain your reasoning

31

10

144

Shaun Ralston

@shaunralston

4 months

@petergyang Love it, @petergyang 🤣 SO grateful for the partnership

Reddit, Inc.

@Reddit

4 months

Reddit and OpenAI today announced a partnership to benefit both the Reddit and OpenAI user communities in a number of ways. Learn more on our corporate blog.

44

114

606

3

1

142

Shaun Ralston

@shaunralston

3 months

Today, GPTs are now available to all FREE users, providing access to 100+ million weekly. And all GPTs are using @OpenAI 's newest flagship model, GPT-4o. Keep on iterating . . .

Introducing GPT-4o and more tools to ChatGPT free users

Introducing GPT-4o and more tools to ChatGPT free users We are launching our newest flagship model and making more capabilities available for free in ChatGPT.

openai.com

7

19

126

Shaun Ralston

@shaunralston

3 months

Blown away 🤯 Real time screen capture below; a 10MB .mp4 (video/audio) @lexfridman excerpt of a @sama interview transcribed to text in less than one second on @GroqInc LPUs. 🚀 Seriously, @GavinSherry @JonathanRoss321 @sundeep , whatever you're feeding #GroqCloud , I want some for

Shaun Ralston

@shaunralston

3 months

Wicked fast Text-to-Speech in the @GroqInc Playground. Thanks @sundeep . Devs, be sure to check out Whisper V3 at #GroqCloud

6

8

54

8

26

125

Shaun Ralston

@shaunralston

1 month

Yes, creepy AF. But, just to be clear, this occured during red teaming, not in the deployed model. Now, only the authorized preset voices speak (no user voice cloning), thanks to 'a classifier' that detects if the output is using an approved voice. #SafetyFirst

Chubby♨️

@kimmonismus

1 month

To be fair, this explains to me a lot why GPT-4o-advanced voice is delayed. Just imagine giving it to your children and then this happens. Because some people don’t get it: after the „NO“ GPT cloned her voice by its own and speaks in her voice. What the.. Edit2: yes, it’s valid

56

55

743

15

6

113

Shaun Ralston

@shaunralston

5 months

Prompt, "photorealistic video with the view from a motorcycle of a futuristic mountain city road at sunset." Another fun #Sora generation.

18

11

109

Shaun Ralston

@shaunralston

6 months

@AiBreakfast This is very old and out of context. @satyanadella said this last year in November amidst the turmoil with Sam's ouster and staff threatening to leave. Satya was trying to reassure shareholders that the huge investment in OpenAI wasn't going to impact Microsoft. Read it again in

5

9

108

Shaun Ralston

@shaunralston

26 days

Spot on, @apples_jimmy 🍎

Jimmy Apples 🍎/acc

@apples_jimmy

27 days

Confirmed. Now let’s never talk about this update ever again.

0

9

181

2

4

107

Shaun Ralston

@shaunralston

1 month

Wow. FLUX.1 Pro on @poe_platform ; best kept secret w/access to best LLMs, image generators, custom bots, coding w/artifact windows; super friendly w/o need for complex integrations or multiple subs. Great work @adamdangelo ~ Poe is lit 🔥

8

10

100

Shaun Ralston

@shaunralston

14 days

@nikicaga Just reverse those colors. Trump can win all the swing states and still lose! 😆 What a silly premise - you really think California and New York are going to Trump?

3

1

93

Shaun Ralston

@shaunralston

1 month

Do you think I’m white? Listen to a GPT-4o voice jailbreak (early red teaming) and how it could lead to biased inferences + how this risk is now handled. Details in thread 🧵👇

10

11

84

Shaun Ralston

@shaunralston

14 days

@MaxMaaxMaaax PA is very most likely blue.

3

1

83

Shaun Ralston

@shaunralston

11 days

Witness the dawn of machine speech in 1939 to @OpenAI 's GPT-4o’s advanced voice today, a mind-blowing 85-year journey 🤯 check the exponential rise of human progress in 🧵👇

7

13

87

Shaun Ralston

@shaunralston

22 days

Everyone’s waiting for the next thing, but I’m blown away by what we have today. In under 3 minutes, I prompted a @midjourney v6.1 image + animated w/ @runwayml Gen-3 Alpha. The future is here now, go create!

6

4

76

Shaun Ralston

@shaunralston

3 months

@ylecun Thank you @ylecun - If more people understood that Large Language Models (LLMs) can't reason, many concerns about AGI and robots taking over the world would diminish, leading to a calmer perspective on AI's role.

5

6

74

Shaun Ralston

@shaunralston

5 months

@futuristflower Love this music creation. GPT-5 will not be (anywhere close to) AGI. However, it will knock off socks.

9

5

73

Shaun Ralston

@shaunralston

2 months

Some ChatGPT Plus users experienced GPT-4o advanced voice, for a few minutes; it was mistake (now corrected). Tom's Guide AI Editor @RyanMorrisonJer explains what happened. Great reporting.

GPT-4o advanced voice 'accidentally' leaked out to some users — here's what happened

It can generate sound effects

www.tomsguide.com

4

12

67

Shaun Ralston

@shaunralston

2 months

@skirano Pietro, this release is about price/performance. GPT-3.5 was no longer competitive against Gemini Flash, Claude Haiku, etc. Many business customers using LLM APIs for routine functions (company chatbots, standard work), need/want low token costs. GPT-4o mini sets a new standard

10

0

69

Shaun Ralston

@shaunralston

4 months

Timing feels legit. 🚀 Exciting.

nic

@nicdunz

4 months

Expect GPT-4o voice mode to begin a slow alpha roll out beginning June 3. This could change.

13

151

2

3

65

Shaun Ralston

@shaunralston

6 months

Happy 🍀 Day! "An elephant made of leaves running in the jungle." Another fun (and green) #Sora generation @MattVidPro

4

11

63

Shaun Ralston

@shaunralston

24 days

Seems fitting that 4o models rank top 4 in @lmsysorg

5

1

64

Shaun Ralston

@shaunralston

10 months

@npew Thanks @nepew . The Q-Anon people, along with @Reuters , will have a field day. 🤣 Q-learning being adept at arithmetic is one logical component of moving us closer, but we need many components for AGI. I predict we'll have a thousand "doomsayer" posts by Thanksgiving morning.

1

2

63

Shaun Ralston

@shaunralston

6 months

@minchoi Great model, no where near AGI (definition dependent). Some people used to call GPT 3.5 as super human. None of the current models are AGI.

8

0

62

Shaun Ralston

@shaunralston

27 days

Wondering about Q*, STaR, STRawbery, and the ‘sus-column-r’ model in @lmsysorg ? Check this rubric test by @matthewberman 🎥 (be sure to follow him, too) 👉 video at 🚀 More on how models may be improving in 🧵👇

4

3

59

Shaun Ralston

@shaunralston

2 months

@Scobleizer That's because your posts are positive and forward thinking. Michele's post is critical and frustrating. Keep your wonderful demeanor, Robert.

2

1

61

Shaun Ralston

@shaunralston

1 month

@growing_daniel Look for the Adam's apple, @growing_daniel

1

60

Shaun Ralston

@shaunralston

2 months

Devs! GPT-4o mini, API on the cheap!

OpenAI unveils GPT-4o mini, a smaller and cheaper AI model | TechCrunch

OpenAI introduced GPT-4o mini on Thursday, its latest small AI model. The company says GPT-4o mini, which is cheaper and faster than OpenAI's current

techcrunch.com

6

8

61

Shaun Ralston

@shaunralston

5 months

Llama on Speed! 🦙💨 @GroqInc just started running @AIatMeta Llama3-70b; 🚀 wicked fast via the LPU inference engine. Great job 👏 @sundeep . A must try, visit

2

9

61

Shaun Ralston

@shaunralston

1 month

Powered by @OpenAI

Figure

@Figure_robot

1 month

Figure 02 coming 8/6

180

786

4K

4

3

60

Shaun Ralston

@shaunralston

1 month

Check out today's @OpenAI report outlining safety work carried out before releasing GPT-4o #SystemCard

GPT-4o System Card

This report outlines the safety work carried out prior to releasing GPT-4o including external red teaming, frontier risk evaluations according to our Preparedness Framework, and an overview of the...

openai.com

1

5

58

Shaun Ralston

@shaunralston

9 months

@chaseleantj @chaseleantj , Incredible - finding that when left to creativity (minimal prompt instructions), v6 is more focused than v5.2. Below prompt: A pecan paradise (left v5.2, right v6). Love UR examples + categories.

3

1

55

Shaun Ralston

@shaunralston

6 months

@futuristflower At this point, these are just numbers - margin of error. Various models excel at various tasks. At this point, use-case, speed and pricing should be what dictate choice . . . not a .5% variance on some random benchmark. For example, Inflection 2.5 is superior for human chat

1

55

Shaun Ralston

@shaunralston

11 days

let the pre-release testing roll

Sam Altman

@sama

11 days

we are happy to have reached an agreement with the US AI Safety Institute for pre-release testing of our future models. for many reasons, we think it's important that this happens at the national level. US needs to continue to lead!

2K

1K

12K

3

2

56

Shaun Ralston

@shaunralston

3 months

Wicked fast Text-to-Speech in the @GroqInc Playground. Thanks @sundeep . Devs, be sure to check out Whisper V3 at #GroqCloud

6

8

54

Shaun Ralston

@shaunralston

15 days

@flowersslop Advanced voice mode is fun, but doesn't really add intelligence. Most functions, absent the playful voice inflections, are available right now with GPT-4o standard voice (TTS, Whisper).

4

0

53

Shaun Ralston

@shaunralston

29 days

Amen, brother!

Logan Kilpatrick

@OfficialLoganK

29 days

Anyone who thinks @iruletheworldmo is AGI or some advanced model has no idea how OpenAI operates as a company.

94

32

1K

3

0

54

Shaun Ralston

@shaunralston

4 months

Devs! Check out GitHub for evals comparing GPT-4 models. Yes, GPT-4o is EPIC, but GPT-4-Turbo bests on complex reasoning and arithmetic (DROP dataset, anyone?),

4

2

51

Shaun Ralston

@shaunralston

9 months

@chaseleantj Love @midjourney v6 ALPHA prompt adherence, but its REAL power (imho) is MJ's magical ability roam creative. My simple prompt, "surreal highly detailed photo so complex no human can recreate it." Uniquely MJ output.

2

1

49

Shaun Ralston

@shaunralston

1 month

@acnewsitics

12

13

50

Shaun Ralston

@shaunralston

1 month

@gtconway3d George, I'm dying. You made my day.

0

1

50

Shaun Ralston

@shaunralston

13 days

Thanks @steph_palazzolo ! Great 411 on fast-tracking “Strawberry,” which excels in complex reasoning and problem-solving to stay ahead in the competitive AI landscape, a key to enhance ChatGPT and train future models like Orion.

OpenAI Races to Launch ‘Strawberry’ Reasoning AI to Boost Chatbot Business

As OpenAI looks to raise more capital, its researchers are trying to launch a new artificial intelligence product they believe can reason through tough problems much better than its existing AI....

www.theinformation.com

4

48

Shaun Ralston

@shaunralston

2 months

@ScottJenningsKY You can still delete this post.

2

0

46

Shaun Ralston

@shaunralston

22 days

DEVS miss @OfficialLoganK support at OAI.

Logan Kilpatrick

@OfficialLoganK

22 days

@AlpinDale @AnthropicAI @OpenAI On it

8

5

115

2

0

47

Shaun Ralston

@shaunralston

1 month

Wow, Ammaar. Nobody knows more about audio than @elevenlabsio 's chief designer. Check his push of GPT-4o advanced voice into fun multi-modal. Bravo!

Ammaar Reshi

@ammaar

1 month

ChatGPT Advanced Voice Mode can generate sound effects while it's talking! 🔊 Watch it give me a weather report while creating the sounds for thunder, rain, and howling wind! ⛈️ Such a fun use of multi-modal audio.

9

13

127

3

2

46

Shaun Ralston

@shaunralston

30 days

Abstract by @Stanford researchers @ericzelikman , @gharik , @EchoShao8899 , @varunajayasiri , @nickhaber , and @noahdgoodman examining how LLMs can learn to generate rationales at each token to explain future text, improving their predictions, full PDF at

1

5

46

Shaun Ralston

@shaunralston

1 month

Your are right, Grandmaster @aidan_mclau , but it won't suck ♟️ think amplified RL, combined with new self-taught abilities (ya, tabula rasa), to exponentially enhance general intelligence; ofc the future isn’t just in scaling—it’s in evolving how we teach it 🧠 Videbimus.

5

3

44

Shaun Ralston

@shaunralston

6 months

Happy Friday! "A wide shot video of a rally car driving through a redwood forest and showing up behind the trees, spraying dirt into the air as it drifts around a sharp turn, cinematic film shot in 35mm." Another fun #Sora generation. @MattVidPro

9

6

44

Shaun Ralston

@shaunralston

29 days

@chatgpt21 Yes, in the latest iteration, there are some improvements. Bit it's important to understand that LLMs process text by breaking it down into tokens (see pic), which are chunks of characters that can range from single letters to entire words. These tokens are the basic units the

4

2

44

Shaun Ralston

@shaunralston

25 days

Gen-3 Alpha Turbo, blazing new speed; just generated this 10 second video in 43 seconds ⚡️🎥 Thank you @runwayml , available now 🚀

Runway

@runwayml

25 days

Gen-3 Alpha Turbo Image to Video is now available and can generate 7x faster for half the price of the original Gen-3 Alpha. All while still matching performance across many use cases. Turbo is available for all plans, including trial for free users. More improvements to the

69

111

619

3

1

45

Shaun Ralston

@shaunralston

2 months

Nothing but love to the @GroqInc team ❤️ Not only is your inference engine lightning speed (as always) ⚡️ but so is your new model deployment 🚀 #405b

3

1

43

Shaun Ralston

@shaunralston

1 month

Devs! 🍓🍓 distracting you? @MistralAI Agents are here now; develop and share generative AI apps + workflows in Mistral Large 2. Super easy. Dive in and build, baby, build.

2

8

43

Shaun Ralston

@shaunralston

6 months

@bitcloud Ha ha ha. The LLM is just following your directions after the comma. You can try it with ANYTHING. Sorry, but Claude can NOT think; it can just follow orders.

5

2

41

Shaun Ralston

@shaunralston

3 months

. @Apple 's 1987 "Knowledge Navigator" video predicted unthinkable tech we use today; tablets, touch screens, voice, video, and AI agents. @Scobleizer @BrianRoemmele ~ watch how Apple's decades old vision manifest today. #ThinkDifferent

3

9

42

Shaun Ralston

@shaunralston

4 months

Mesmerizing 75 original OpenAI #Sora videos, set to Johann Strauss’ iconic “Blue Danube!” The magic of AI generation motion + classical music combined, compiled by @1BlueSpork – bravo!

2

8

42

Shaun Ralston

@shaunralston

5 months

@mattshumer_ @GroqInc 100% @mattshumer_ ~ I keep refreshing in disbelief.

0

2

42

Shaun Ralston

@shaunralston

2 months

Just a day after @AIatMeta Llama-3-400b release, @MistralAI 's new model, Mistral Large 2, launches spectacular performance proving that OSS seriously rivals the top AI giants! Feel the e/acc 🚀 Visit

4

5

41

Shaun Ralston

@shaunralston

3 months

E/acc congrats @elon @SpaceX for today's successful #Starship takeoff + splashdown! 🌟 Groundbreaking achievement; the reusability and massive payload capacity will revolutionize space travel to Mars, the Moon, and beyond. 🚀

0

3

42

Shaun Ralston

@shaunralston

1 month

@harris_wins

17

9

40

Shaun Ralston

@shaunralston

3 months

@ai_for_success Just use Microsoft CoPilot (it's GPT-4o) and working. Unexpected traffic on those other servers likely causing issues, but Microsoft owns the Azure cloud - viturally unlimited compute.

Microsoft Copilot: Your everyday AI companion

Microsoft Copilot leverages the power of AI to boost productivity, unlock creativity, and helps you understand information better with a simple chat experience.

copilot.microsoft.com

4

1

40

Shaun Ralston

@shaunralston

1 month

Echoing @TheRealAdamG ~ Stop waiting for ‘advanced voice’ or GPT-5. You have revolutionary #AI at your fingertips. Iterations will always be forthcoming, but you hold the power NOW, see thread 🧵👇

7

5

40

Shaun Ralston

@shaunralston

1 month

@acnewsitics

7

6

39

Shaun Ralston

@shaunralston

8 months

@osanseviero @MistralAI @kyutai_labs @AIatMeta @GoogleDeepMind Very cool, indeed. @OpenAI also contemplating making France as EU headquarters.

Digital: OpenAI has its sights set on France as the location for its European HQ - Business...

Sam Altman, the CEO of OpenAI — the company behind the wildly popular ChatGPT application — is considering setting up its European HQ in France. The CEO of OpenAI disclosed the main reason for...

www.business-solutions-atlantic-france.com

1

4

38

Shaun Ralston

@shaunralston

3 months

@itsandrewgao @OpenAI Very impressive, indeed. I wonder how GPT-4o will handle my grandmother's famous Chihuahua muffins? 🤣

1

0

38

Shaun Ralston

@shaunralston

3 months

Sometimes, shipping is the real MVP. Today @runwayml announced Gen-3 Alpha (video). Like #Sora , #Veo , and #KlingAI , G3A is not yet in public release. Hat tip to @LumaLabsAI #DreamMachine and @gravicle for getting yours out the door 🚀📦💪 Brava!

0

3

37

Shaun Ralston

@shaunralston

3 months

We're so back (again).

2

0

38

Shaun Ralston

@shaunralston

17 days

"A kitten wearing a Chinese Monkey King costume holds a golden stick weapon and is walking across a weathered roof"; @midjourney 6.1 + @runwayml Gen-3 Alpha.

4

2

36

Shaun Ralston

@shaunralston

4 months

Watch @satyanadella Microsoft's 'CoPilot everything' keynote on Tuesday; a MacBook killer AI-powered (Snapdragon X Elite) Surface Pro + Laptop. Also, expecting a surprise @OpenAI 'leadership' appearance.

Microsoft Build 2024: Day 1 #MSBuild

For the full event experience, please head to ➡️ https://aka.ms/MSBuild2024EventJoin us for Day 1 of Microsoft Build 2024 Microsoft Build is for developers, ...

www.youtube.com

3

4

36

Shaun Ralston

@shaunralston

1 month

@Scobleizer Robert, with the exception of a lot of bored people posting fruit, nothing is happening tonight.

7

1

37

Shaun Ralston

@shaunralston

2 months

Who has been playing with the @AIatMeta Llama3-405b preview this morning?

4

1

36

Shaun Ralston

@shaunralston

3 months

Looks like even #AI needs to take a mental health break. #ChatGPTdown #AIburnout 🤖😴

6

1

34

Shaun Ralston

@shaunralston

5 months

Best marketing (and engineering) ever, @Humane ! Now, that's how you launch; congrats @bella_bongiorno , @imranchaudhri , @Jose_Cong , @jaredostdiek , @JArnost , and others. #EclipsingTheCompetition

0

1

34

Shaun Ralston

@shaunralston

1 month

@bindureddy Yes, #AGI achieved; Anecdotes, Gossip, Innuendo 😉

0

34

Shaun Ralston

@shaunralston

5 months

Happy #Sora Sunday, “a first-person perspective of traveling through a tunnel made of mirrors reflecting a starry galaxy, creating an illusion of infinite depth and motion.” Another wild #Sora generation.

2

8

34

Shaun Ralston

@shaunralston

6 months

. @rowancheung is a reliable, calm, and trusted voice. His assessment and chronology here is spot on. If you want to cut through the hype and understand what’s really going on, be sure to read this thread. Nice job, Rowan!

Rowan Cheung

@rowancheung

6 months

So, Elon is suing OpenAI partly because he believes OpenAI's Q* model has a strong claim to being AGI. Q* is OpenAI's secret breakthrough that was leaked during Sam Altman's ousting in November. It's mostly still unknown, but here’s everything we know about Q* so far:

185

564

5K

0

1

34

Shaun Ralston

@shaunralston

21 days

“Toto, I've a feeling we're not in Kansas anymore.” -- @MatthewBerman @datadoghq 🧵👇

4

0

33

Shaun Ralston

@shaunralston

30 days

@BrianRoemmele AI is getting really good . . . almost convincing.

1

0

33

Shaun Ralston

@shaunralston

1 month

Toss 'em in the blender; together with other fruits and protein, they make great smoothies.

Sam Altman

@sama

1 month

i love summer in the garden

1K

793

13K

6

2

33

Shaun Ralston

@shaunralston

1 month

🚀 Winner Alert, Must Watch! 🎉 @MistralAI Large 2 outshines Claude-3.5 Sonnet and Llama-3-405b! 🏆✨ Leading the @MatthewBerman rubric. 🥇 Exceptional coding, razor-sharp replies, and top-notch moral judgment! 💪🔥

Mistral Large 2 | INSANE Model Overshadowed by LLaMA 405b (Fully...

Here's a full test of Mistral Large 2, an absolutely phenomenal model that dropped around the same time as LLaMA 3.1 405b, so it didn't get nearly the same r...

www.youtube.com

1

5

33

Shaun Ralston

@shaunralston

3 months

@ID_AA_Carmack @SpaceX John, this moment made me cry. Thanks for sharing @ID_AA_Carmack and @elonmusk . Future hope!